Quality of Control Arms in Randomized Trials of Chronic Lymphocytic Leukemia Enrolling in the USA: A Systematic Review

Introduction: Many ongoing clinical trials in chronic lymphocytic leukemia (CLL) are investigating "chemotherapy-free” combinations of novel agents administered in a time-limited manner both in the frontline and relapsed/refractory setting. However, there is a concern that the results of trials testing these novel treatment combinations can be difficult to interpret if the therapy in the control groups is suboptimal and/or are not commonly used in the USA. A suboptimal control group reduces the trial's external validity as it introduces a bias in favor of the experimental arm. To date, an empirical evaluation of the quality of control groups in randomized trials of chronic lymphocytic leukemia has not been undertaken. We sought to carry out a systematic review of randomized trials of CLL between Jan 2010 and Dec 2021 to establish the quality of the control group therapies compared to USA standards as determined by NCCN and expert reviews.

Methods: PubMed (MEDLINE), Embase, Cochrane, Scopus, and Web of Science were searched for articles of randomized trials of CLL that enrolled participants in the USA between January 1, 2010, to December 31, 2021. Standard-of-care therapy for the USA was determined by evaluating the literature (expert reviews) and published guidelines (mainly NCCN) prior to the start of trial enrollment. The percentage of randomized trials that used suboptimal control arms was then calculated. A control group therapy was considered suboptimal if the choice of control arm was restricted (i.e., excluded an active therapy), and/or if prior randomized trials had shown the control agent (or class) was inferior to an available alternative or by limiting combinations in the control group.

Results: A total of 1775 articles with our initial search strategy. 625 records were duplicates and were excluded. We screened 1150 articles in our primary screening. Of these, 156 were excluded due to being non-CLL studies, 117 records were excluded due to being duplicates, abstracts, trials in progress, and subgroup analysis, 258 records were excluded due to being review articles, and case series/reports, 564 records were excluded due to being single-arm studies. 55 articles were eligible for secondary screening. Of these, 29 did not meet inclusion criteria due to not enrolling in the U.S and 7 were excluded due to enrollment start date before Jan 2010. 19 trails were eligible for the final analysis. Of the 19 trials that were assessed in the final analysis, we identified 10 (52%) trials that enrolled patients in a suboptimal control arm. When classified further by the nature of the suboptimal control arm, 7 (36%) trials omitted active treatment in the control arm by using a control agent known to be inferior to other available agents or not allowing combinations, and 3 (16%) trials omitted active treatment in control arm by limiting investigator's choice.

Conclusion: Our findings show that half of randomized trials of CLL use a suboptimal control group. Although limiting investigator choice is common within the context of a randomized trial and may not be detrimental, approx. a third of randomized trials use control agents that are known to be inferior to other available agents. The choice of control arm of trials enrolling in the USA should be optimized to ensure that it matches what most clinicians would prescribe outside a clinical trial setting.

No relevant conflicts of interest to declare.